Expression and purification of FL S(P2) and RBD proteins

In brief, protein sequences of the Omicron XBB.1.5 sublineage and WT (Wuhan-Hu-1) FL S(P2) encoded by BNT162b2 were used to generate a construct containing a C-terminal TwinStrep tag to facilitate affinity purification and were cloned into a pcDNA3.1(+) vector for expression. The RBD of each FL S(P2) (Omicron XBB.1.5 and WT) was expressed as secreted protein and purified via the engineered C-terminal affinity tag. Both FL S and RBD protein expressions were conducted in Expi293F cells (ThermoFisher Scientific) grown in Expi293 medium.

Expression and purification of FL S(P2)

Expression of proteins was carried out in Expi293F cells (Thermo Fisher Scientific) grown in Expi293 medium. Cells were transiently transfected with S or RBD protein expression constructs in the pcDNA3.1(+) vector. Expression was conducted at 37 °C for 24 hours before adding Expifectamine enhancers (Thermo Fisher Scientific). After addition of enhancers, the temperature was dropped to 32° C and expression was allowed for another 48–72 hours before collecting. A modified protocol of procedures described by Zhang et al.50 was used for purification of the SARS-CoV-2 FL S(P2). Briefly, the transfected cells were lysed in a solution containing Buffer A (100 mM HEPES pH 8.0, 150 mM NaCl, 1 mM EDTA), 1% (w/v) n-dodecyl-β-D-maltopyranoside (DDM, Anatrace), EDTA-free complete protease inhibitor cocktail (Roche), and Pierce Universal Nuclease (Thermo Fisher) at 4 °C for 1 h. After a clarifying spin at 40,000 ´ g for 45 min, the supernatant was filtered with 0.2 mm filter (Nalgene 78018-24, 1 LL) before batch bound onto StrepTactin HP resin (Cytiva) equilibrated with the lysis buffer at 4 °C for 1 h. Resin was collected by centrifugation at 1000 ´ g and loaded onto EconoColumn (Bio-Rad) for gravity flow purification. The column was washed with Buffer A containing 0.5% DDM, 10 mM ATP, and 10 mM MgCl2, followed by additional washes with Buffer A and gradually reduced concentrations of DDM (0.5–0.02%). FL S(P2) was eluted with Buffer A containing 0.02% DDM and 5 mM d-Desthiobiotin. The protein was further purified by size exclusion chromatography (SEC) on a Superose 6 10/300 column (Cytiva) in a buffer containing 25 mM Tris pH 7.5, 150 mM NaCl, 1 mM EDTA, and 0.02% DDM. DDM-purified FL S(P2) was eluted as a single peak over SEC. FL S(P2) protein from the SEC peak fractions were analyzed by denaturing PAGE using a 4–15% Criterion TGX Stain-Free Gel (Bio-Rad, Supplementary Fig. 1), and used in thermostability (Tm), biolayer interferometry (BLI), mass spectrometry and cryogenic electron microscopy (cryo-EM) experiments.

Expression and purification of RBD

The RBDs were expressed using Expi293F cells (Thermo Fisher Scientific) grown in Expi293 medium transiently transfected with the RBD expression constructs in pcDNA3.1(+) vector. The RBD constructs contain an N-terminal S protein leader peptide and coding regions from 324–531 (Omicron XBB.1.5) and 327–528 (WT ancestral strain), respectively, followed by a C-terminal affinity tag as indicated in Supplementary Table 1. Expression was conducted at 37 °C for 120 hours before the proteins were collected from cell culture medium. The affinity tagged RBDs were purified on affinity purification columns first, subsequently on Superdex200 gel filtration column (Cytiva), and stored in a buffer containing 100 mM Tris pH7.5, 150 mM NaCl, and 10% glycerol. RBD was expressed as secreted protein and purified via the engineered C-terminal affinity tag.

Stability of WT and Omicron XBB.1.5 FL S(P2) by Thermal Shift Assay (TSA)

Stability of FL S(P2) proteins was measured by Tycho NT.6 (NanoTemper, firmware version: 1.10.3) and the data were analyzed by the Tycho NT.6 software (version: 1.3.2.880). In brief, a 10 mL solution containing 0.35 mg/mL of protein was loaded into a capillary tube and the ratio of tryptophan fluorescence at emission wavelengths of 350 nm over 330 nm and was measured while ramping the temperature from 35 °C to 95 °C using the pre-programmed protocol of the instrument. The inflection temperature for each thermal melting curve was reported by the Tycho NT.6 software.

Binding kinetics of purified FL S(P2) protein and RBD to immobilized human ACE-2-PD

FL S(P2), with a C-terminal TwinStrep tag expressed in Expi293F cells, was detergent solubilized and purified by affinity and size exclusion chromatography. The peak fraction of the purified FL S(P2) and isolated RBD proteins of Omicron XBB.1.5 and WT strains were assessed by biolayer interferometry (BLI) binding to immobilized human ACE-2-PD on an Octet RED384 (FortéBio) at 25 °C in a running buffer that comprised 25 mM Tris pH 7.5, 150 mM NaCl, 1 mM EDTA, and 0.02% DDM, identical to the protein purification buffers. The highest concentration assessed for both FL S(P2) and RBD was 300 nM, with three additional three-fold dilutions. BLI data were collected with Octet Data Acquisition software (version 10.0.0.87) and processed and analyzed using FortéBio Data Analysis software (version 10.0). Binding curves were reference subtracted and fit to a 1:1 Langmuir model to determine binding kinetics and affinity.

Cryo-EM of Omicron XBB.1.5 FL S(P2)

Purified Omicron XBB.1.5 FL S(P2) was applied to glow discharged Quantifoil R1.2/1.3 200 mesh gold grids and blotted using a Vitrobot Mark IV (ThermoFisher Scientific) before being plunged into liquid ethane cooled by liquid nitrogen. Datasets were collected and analyzed as depicted in Supplementary Fig. 3. The purified sample in DDM at 5.0 mg/mL were applied onto the glow discharged Quantifoil R1.2/1.3 200 mesh gold grids and blotted using a Vitrobot Mark IV (ThermoFisher Scientific). A data set of 6690 movies was recorded using EPU from a Titan Krios G2 transmission electron microscope operating at 300 keV equipped with a Falcon 4i direct electron detector and Selectris Energy Filter (ThermoFisher Scientific). Each movie was collected in counting mode with a pixel size of 0.75 Å/pixel, 10 eV slit and a defocus range of −0.6 μm to −2.6 μm for a total dose of 40.0 e−/Å2. Each data set was imported and processed in CryoSPARC v4.2.1. All movies were adjusted with patch motion correction and patch CTF estimation. Templates were generated from 2D class averages after automative particle picking by blob picker. These 2D class average templates were used for template-based autopicking to pick particles for the rest of the data processing.

From Template Picker, 1,862,402 particles were autopicked and extracted with a box size of 540 pixels. Iterative 2D classification were carried out to select particles with high resolution views of the spike protein (3131,229 particles). Three initial models were generated using all 131,229 particles in ab initio reconstruction resulting in only one map containing 91,663 particles that resemble a spike protein with 1-RBD-up. Heterogeneous refinement of the selected particles resulted in only 1-RBD-up structures. Therefore, all 91,663 particles were subjected to homogeneous refinement, followed by non-uniform refinements, which gave the final 1-RBD-up structure with an overall resolution of 2.98 Å. All refinement steps were done with C1 symmetry. The final resolution was calculated from the Fourier Shell Correlation (FSC) curve from the resolution at the 0.143 FSC cutoff.

A model of the Omicron spike protein structure (PDB: 7TGW)51 was fitted into the final cryo-EM structure and was used as a guide for modeling. The atomic model was built in COOT52 and refined using Phenix real space refinement53. The EM density for the RBD in the up position was weakly resolved. Therefore, the RBD was docked into the EM density and rigid body fitted without side chains unless there were clear side chain densities. The final model including the RBDs were refined in Phenix (version 1.20-4459-0000).

Mass spectrometry characterization of N-linked glycosylation

Mapping of N-linked glycosylation sites was conducted on recombinant purified Omicron XBB.1.5 S(P2) following precipitation using ice cold acetone and incubated overnight at −20 °C. The protein was pelleted, dissolved in 8 M urea, and reduced and alkylated prior to proteolytic digestion. Protein samples were digested in three batches using either trypsin, trypsin/Glu-C, or chymotrypsin. Digested peptide pools from all three reactions were subjected to mass spectrometry analysis to achieve a desired sequence coverage (~92%). Peptides were separated from remaining enzymes using a Microcon-10kDa centrifugal filter (MRCPRT010). The supernatant was collected and lyophilized to dryness. For N-linked analysis, digested samples were reconstituted in O18 water. PNGase F and O-glycosidase were added to remove N-linked and O-linked glycosylation.

The treated digests were analyzed on a Thermo QExactive Oribitrap Mass Spectrometer with an EZ-NanoSpray Source and an EZ-nLC 1200. The peptides were chromatographically separated prior to in-line mass spectrometry analysis with a flow rate of 2 mL/min. The samples were also analyzed on a Thermo Fusion Tribrid Mass Spectrometer outfitted with an EZ-nLC 1200 to perform peptide separations. The system was operated in direct injection mode and the peptides were chromatographically separated prior to in-line mass spectrometry analysis with a flow rate of 450 nL/min. Matrix Science MASCOT and Thermo Freestyle were used for data analysis. The N-linked data was searched with the following variable modifications: Deamidated (NQ), Deamidated:18 O(1) (NQ), HexNAc (N), Oxidation (M).

Animal ethics

All murine experiments were performed at Pfizer, Inc. (Pearl River, NY, USA), which is accredited by the Association for Assessment and Accreditation of Laboratory Animal

Care (AAALAC). All procedures performed on animals were in accordance with regulations and established guidelines and were reviewed and approved by an Institutional Animal Care and Use Committee or through an ethical review process.

BNT162b2 mRNA XBB.1.5 vaccine modification and formulation

The XBB.1.5 adapted vaccine encodes the S(P2) of XBB.1.5 (GISAID EPI_ISL_16292655) on the BNT162b2 RNA backbone. Purified nucleoside-modified RNA was formulated into lipid nanoparticles by mixing together an organic phase lipid mixture with an RNA aqueous phase, and subsequently purifying the mix to yield a lipid nanoparticle composition similar to one previously described54.

Immunogenicity in BNT162b2-experienced mice

Female BALB/c mice (10 per group, age 6–8 weeks; Jackson Laboratory) were vaccinated and bled as shown in Supplementary Fig. 4A. In brief, mice were vaccinated intramuscularly with a 2-dose series (Day 0, 21) of a 0.5 µg dose level of original BNT162b2 WT vaccine, followed by a 3rd dose booster (Day 105) of bivalent WT + Omicron BA.4/5 vaccine, and a 4th dose booster (Day 134) of either monovalent Omicron XBB.1.5, monovalent Omicron BA.4/5, bivalent Omicron XBB.1.5 + BA.4/5 or bivalent WT + Omicron BA.4/5 sublineage-modified vaccines. Bivalent formulations contained equal quantities of each mRNA (0.25 µg each) and a total dose level of 0.5 µg. A control group of ten mice received saline injections according to the same schedule in place of active vaccines. A total volume of 50 µL of vaccine or saline was administered intramuscularly to the upper outer hind leg for each animal. Animals and injection sites were observed immediately after vaccination. Sera were collected for evaluation of pseudovirus neutralizing antibody responses prior to the 4th dose (Day 134) and at the final post-vaccination timepoint (Day 160). Spleens were also collected at Day 160 to evaluate cell-mediated immune responses.

Immunogenicity in naive mice

Female BALB/c mice (10 per group, age 10–12 weeks; Jackson Laboratory) were vaccinated and bled according to the illustration in Supplementary Fig. 4B. In brief, mice were vaccinated intramuscularly on Days 0 and 21 with either monovalent Omicron XBB.1.5, monovalent Omicron BA.4/5, bivalent Omicron XBB.1.5 + BA.4/5 or bivalent WT + Omicron BA.4/5-adapted vaccines. The control group received saline injections according to the same schedule as active vaccine groups. Sera and spleens were collected 28 days after the second dose (day 49) for evaluation of pseudovirus neutralizing antibody responses and cell-mediated immune responses, respectively.

Pseudovirus neutralization assay

Pseudovirus stocks were generated in HEK-293T cells (ATCC, ref.# CRL-3216) using SARS-CoV-2 spike plasmid DNA and vesicular stomatitis virus (VSV; VSVΔG(G)-GFP virus: Kerafast, ref.# EH1019-PM). Serial dilutions of heat-inactivated murine sera (3-fold) were incubated with pseudovirus (VSVΔG(G)-GFP expressing SARS-CoV-2 S protein) for 1 h at 37 °C before inoculating confluent Vero (ATCC, ref.# CCL81.2) cell monolayers in 96-well plates. Fluorescent virus-infected foci were detected 19–21 h after inoculation with an anti–VSV pAb (Imanis Life Sciences, ref# REA005) and Alexa488-conjugated secondary antibody (Invitrogen, ref# A-11008) and enumerated using a CTL Immunospot Analyzer (Cellular Technology Limited). A 50% neutralization titer (NT50) was calculated as the last reciprocal serum dilution at which 50% of the virus is neutralized compared to wells containing virus only. Each serum sample dilution was tested in duplicate. The assay titer range was 20 to 43,740. Any serum sample that yielded a titer >43,470 was prediluted and repeated to extend the upper titer limit; sera that failed to neutralize at the lowest serum dilution (1:20) were reported to have a neutralizing titer of 20 (lower limit of detection, LLOD). VSV-based pseudoviruses used in the assay expressed the S protein from the following SARS-CoV-2 variants: WT (Wuhan-Hu-1, ancestral strain), BA.4/5, XBB.1.5, XBB.1.16, XBB.1.16.1, XBB.2.3, EG.5.1, HV.1 and BA.2.86. Amino acid sequence alignments for all tested pseudoviruses are provided in Supplementary Fig. 8.

T-cell response assay

In brief, antigen-specific T cell responses were analyzed from murine splenocytes with a flow cytometry-based intracellular cytokine staining (ICS) assay, comparing unstimulated (DMSO) response to those observed in splenocytes after stimulation with a peptide library. Freshly-isolated splenocytes (2 × 106 cells/well) were cultured in cRPMI with media containing DMSO only (unstimulated) or specific amino acid (aa) peptide libraries (15aa, 11aa overlap, 1 to 2 µg/mL/peptide) representing the S amino acid sequences of the original SARS-CoV-2 Wuhan strain (WT), Omicron BA.4/5, and Omicron XBB.1.5 sublineages, separately (JPT, catalog #s PM-SARS2-SMUT10-2, PM-SARS2-SMUT15-1, PM-WCPV-S-1), for 5 h at 37 °C in the presence of protein transport inhibitors, GolgiPlug and GolgiStop. Following stimulation, splenocytes were incubated with fluorescent-conjugated antibodies to the surface proteins CD19, CD3, CD4 and CD8 (25 ± 5 min at RT) followed by fixation and permeabilization and staining for CD154 (CD40L), IFN-γ, TNF-α, IL-2, IL-4, and IL-10 (25 ± 5 min at RT). Ebioscience fixable viability dye eFluor 506 was used exclude dead cells. After staining, the cells were washed and resuspended in flow cytometry (FC) buffer (2% FBS in PBS). Samples were acquired on a BD LSR Fortessa flow cytometer with BD FACSDiva software. Acquired data files were analyzed using BD FlowJo™ (version 10.8). Results are background (media-DMSO) subtracted and shown as percentage of CD154+ cytokine-expressing CD4+ T cells and CD8+ T cells. The Tcell gating strategy is shown in Supplementary Fig. 7.

Animal blood collection and splenocyte isolation

For interim blood draws, mice were bled via the submandibular route. Approximately 150 μL of whole blood was collected dropwise directly into microtainer tubes containing serum separators. For terminal blood draws, the entire available blood volume was collected via cardiac puncture. At each study end, mice were euthanized under a surgical plane of isoflurane and cervical dislocation was performed as a secondary method to confirm death. Submandibular bleeds and vaccinations were not done under sedation. At all blood collection time points, blood tubes remained at room temperature (RT) for at least 30 min prior to centrifuging at 12,300 ´ g for 3 min. Each serum sample was aliquoted and heat inactivated at 56 °C for 30 min. Samples were stored at −80 °C after testing.

For flow cytometry of murine splenocytes, spleens were collected from five mice per group at the final time point for each study. Spleens were placed in a 100 µm cell strainer (BD Falcon) immersed in 7 mL of complete RPMI (cRPMI: 10% FBS/RPMI; Pen-Strep; Sodium pyruvate) per mouse per well of a 6-well plate. The plates were maintained on ice during transit and before processing for single cell suspension. Spleens were homogenized, subjected to RBC lysis, and passaged through a cell strainer to remove RBCs and clumps.

Statistical analysis

Mouse immunogenicity data were analyzed using SAS version 9.4. All statistical analyses were performed using ANOVA on log-transformed data. Comparisons were made on mouse sera across pseudoviruses at the last post-vaccination timepoint of each study with Dunnett’s test for multiple comparisons. For intergroup comparisons, the bivalent WT + Omicron BA.4/5 vaccine group was the reference; for intragroup comparisons (pseudoviruses within a vaccine group), the Omicron XBB.1.5 was the reference. All tests were two-tailed. A p-value of less than 0.05 was considered statistically significant.